Profiling and Imaging Sonar Fusion Based 3D Normal Distribution Transform Mapping for AUV Application

To expand operation coverage and period of time with low cost, self-localization is needed for long range AUVs. For this, AUVs need to perceive the surrounding environment and localize its position based on the perceived information. The types of sensors that can be used for sensing environment are extremely limited in underwater because of water turbidity and low illuminance. Sonars are a conventional solution for underwater sensing, widely used for sensing underwater environment. Sonars, however, have perceptual ambiguity depending on viewpoint, and the loss of elevation information, which makes difficult to use. One of the ways to solve the difficulties of sonar is to use plenty of data acquired from varied angles and locations. More simple way is to use multiple sonars. In this study, we propose imaging and profiling sonar based 3D mapping. We presented 3D point cloud from two different sonars, and implemented NDT mapping to combine the data set. Considering the motion characteristic of AUV, we derived 4-DOF NDT algorithm and tested it in indoor water tank. As a result, we obtained consistent 3D point cloud map.